diff --git a/lib/crewai/src/crewai/agent/core.py b/lib/crewai/src/crewai/agent/core.py
index 6c2626a28..9d50dac4b 100644
--- a/lib/crewai/src/crewai/agent/core.py
+++ b/lib/crewai/src/crewai/agent/core.py
@@ -900,6 +900,11 @@ class Agent(BaseAgent):
         self.agent_executor.tools_handler = self.tools_handler
         self.agent_executor.request_within_rpm_limit = rpm_limit_fn
 
+        # Clear accumulated messages and reset iterations from previous tasks
+        # to prevent context pollution when agent is reused across sequential tasks
+        self.agent_executor.messages = []
+        self.agent_executor.iterations = 0
+
         if self.agent_executor.llm:
             existing_stop = getattr(self.agent_executor.llm, "stop", [])
             self.agent_executor.llm.stop = list(
diff --git a/lib/crewai/tests/agents/test_agent.py b/lib/crewai/tests/agents/test_agent.py
index 32130f900..a6c7971ff 100644
--- a/lib/crewai/tests/agents/test_agent.py
+++ b/lib/crewai/tests/agents/test_agent.py
@@ -2351,3 +2351,138 @@ def test_agent_without_apps_no_platform_tools():
 
     tools = crew._prepare_tools(agent, task, [])
     assert tools == []
+
+
+def test_agent_executor_messages_cleared_between_tasks():
+    """Test that agent executor messages are cleared when agent is reused across tasks.
+
+    This test verifies the fix for issue #4319 where messages accumulated when
+    the same agent was reused across multiple sequential tasks, causing:
+    - Duplicate system messages
+    - Context pollution
+    - Eventually crashes with 'Invalid response from LLM call - None or empty'
+    """
+    agent = Agent(
+        role="Test Agent",
+        goal="Execute tasks",
+        backstory="Test agent for message accumulation fix",
+        allow_delegation=False,
+    )
+
+    task1 = Task(
+        description="Task 1: Simple calculation",
+        expected_output="Result",
+        agent=agent,
+    )
+
+    task2 = Task(
+        description="Task 2: Another calculation",
+        expected_output="Result",
+        agent=agent,
+    )
+
+    task3 = Task(
+        description="Task 3: Yet another calculation",
+        expected_output="Result",
+        agent=agent,
+    )
+
+    # Create the agent executor for the first task
+    agent.create_agent_executor(tools=[], task=task1)
+    assert agent.agent_executor is not None
+    initial_executor = agent.agent_executor
+
+    # Simulate messages being added during task execution
+    agent.agent_executor.messages = [
+        {"role": "system", "content": "System prompt"},
+        {"role": "user", "content": "Task 1 prompt"},
+        {"role": "assistant", "content": "Task 1 response"},
+    ]
+    agent.agent_executor.iterations = 5
+
+    # Reuse the agent for task 2 - this should clear messages
+    agent.create_agent_executor(tools=[], task=task2)
+
+    # Verify the same executor instance is reused
+    assert agent.agent_executor is initial_executor
+
+    # Verify messages were cleared
+    assert agent.agent_executor.messages == [], (
+        "Messages should be cleared when agent executor is reused for a new task"
+    )
+
+    # Verify iterations were reset
+    assert agent.agent_executor.iterations == 0, (
+        "Iterations should be reset when agent executor is reused for a new task"
+    )
+
+    # Simulate more messages for task 2
+    agent.agent_executor.messages = [
+        {"role": "system", "content": "System prompt"},
+        {"role": "user", "content": "Task 2 prompt"},
+        {"role": "assistant", "content": "Task 2 response"},
+    ]
+    agent.agent_executor.iterations = 3
+
+    # Reuse the agent for task 3 - this should clear messages again
+    agent.create_agent_executor(tools=[], task=task3)
+
+    # Verify the same executor instance is still reused
+    assert agent.agent_executor is initial_executor
+
+    # Verify messages were cleared again
+    assert agent.agent_executor.messages == [], (
+        "Messages should be cleared each time agent executor is reused"
+    )
+
+    # Verify iterations were reset again
+    assert agent.agent_executor.iterations == 0, (
+        "Iterations should be reset each time agent executor is reused"
+    )
+
+
+def test_agent_executor_state_isolation_across_crews():
+    """Test that agent executor state is properly isolated when same agent is used in multiple crews.
+
+    This is a common pattern in Flow-based workflows where agents are reused
+    across multiple crew kickoffs.
+    """
+    agent = Agent(
+        role="Reusable Agent",
+        goal="Execute tasks across multiple crews",
+        backstory="Agent that gets reused",
+        allow_delegation=False,
+    )
+
+    # First crew with first task
+    task1 = Task(
+        description="First crew task",
+        expected_output="Result 1",
+        agent=agent,
+    )
+
+    # Create executor for first task
+    agent.create_agent_executor(tools=[], task=task1)
+    executor = agent.agent_executor
+
+    # Simulate task execution adding messages
+    executor.messages = [
+        {"role": "system", "content": "System 1"},
+        {"role": "user", "content": "User 1"},
+    ]
+    executor.iterations = 2
+
+    # Second crew with second task (simulating Flow pattern)
+    task2 = Task(
+        description="Second crew task",
+        expected_output="Result 2",
+        agent=agent,
+    )
+
+    # Reuse agent for second task
+    agent.create_agent_executor(tools=[], task=task2)
+
+    # State should be clean for the new task
+    assert executor.messages == []
+    assert executor.iterations == 0
+    assert executor.task == task2